Improve type detection for BaseNode Inputs and Outputs

[coderabbitai]

Problem

Currently, the BaseNode framework requires all Inputs and Outputs fields to be typed as str, which leads to several issues:

1. Type casting required: Developers need to manually cast values (e.g., int(self.inputs.items))
2. Type ignore comments: Multiple # type: ignore comments are needed to suppress type checker warnings
3. Loss of type safety: The framework doesn't leverage Python's type system effectively
4. Poor developer experience: IDE support and autocompletion are limited

Example

Currently we need to write:

class Inputs(BaseModel):
    items: str  # Should be int

async def execute(self):
    count = int(self.inputs.items)  # type: ignore
    return [self.Outputs(processed=str(i)) for i in range(count)]

Instead of the more natural:

class Inputs(BaseModel):
    items: int

async def execute(self):
    count = self.inputs.items
    return [self.Outputs(processed=str(i)) for i in range(count)]

Expected Outcome

The BaseNode framework should:

• Support proper type annotations for Inputs and Outputs fields
• Automatically handle type conversion/validation where needed
• Eliminate the need for manual type casting and type ignore comments
• Provide better IDE support and type safety

Context

This issue was identified during code review of comprehensive test cases for BaseNode.

Related PR: #262
Related Comment: #262 (comment)

cc: @NiveditJain